In response to the self-undermining problem for modest accounts of rational belief, some have proposed that an agent may rationally lose confidence in the truth of these accounts, while continuing to believe as the accounts prescribe. Such agents believe akratically. Many reject the possibility of rational akrasia. Others have defended it—at least in cases where an agent rationally sees her own beliefs as more accurate than rational alternatives would be. This paper argues that akrasia can be rational, but that defending rational akrasia based on an agent’s views about accuracy cannot succeed. Fortunately, however, the defense is not necessary.
